The R&D incentive is making companies afraid to claim it

Small businesses are now more than half the claimants. The audit office found most compliance activity turned up non-compliance, and the bill can arrive years after the money is spent.

Australia's Research and Development Tax Incentive (R&DTI) exists to make risk affordable. Transparency data released by the Australian Taxation Office (ATO) this month shows how far it reaches: 13,490 companies reported A$16.7 billion in research and development expenditure in 2023-24, and small businesses made up 51 per cent of the companies using the scheme, reporting A$3.1 billion between 6,920 of them.

The problem is what happens afterwards. The Australian National Audit Office, reviewing the program's administration, found that between 73 and 76 per cent of the ATO's completed R&DTI compliance activities from 2017-18 to 2019-20 led to the identification of non-compliance. Liabilities raised on routine compliance work ranged from A$70.1 million to A$105.3 million a year across the same period. Those figures describe what happens once a claim is examined, not the odds of being examined. For a company that has already spent the refund, the distinction is academic.

Kris Gale, an independent contractor to R&D tax specialist Michael Johnson Associates, puts it in blunt terms, saying the program has been used as a sword to send companies into bankruptcy. Startups fund development with a refund, then face an order to repay it years later with penalties and interest attached.Anna Wright co-founded BindiMaps, an indoor wayfinding business whose accessibility technology has been used at the Australian Open and NSW Parliament House. 

The company entered voluntary administration in September 2025. Wright says a second ATO review, covering the innovation program and related capital gains tax questions, left an outstanding liability the business could not meet and triggered a cash-flow crunch it could not refinance in time. The business was bought out of administration and trades under new ownership. Directors can remain personally exposed to tax liabilities after a company's assets have been sold.

Underneath the individual cases sits a structural argument about who decides what counts as research. The incentive is jointly administered: the Department of Industry, Science and Resources registers companies and determines whether activities qualify as eligible R&D, while the ATO decides entitlement to the offset and runs compliance. Gale argues the ATO has drifted into making technical judgments about innovation that belong to the department. The ATO rejects that, saying its role is confined to checking that claimed expenses are real and genuinely related to eligible activity.

The rules are also about to move. The government plans to overhaul the incentive from July 2028, increasing support for core R&D activities while removing it for supporting activities that enable the research. That distinction has mattered little until now because both qualified. Gale's view is that fifteen years of case law is about to be reset, with software development and clinical trials the least certain of all.

Claim, but stop treating the refund as revenue. The single behavioural change that protects a small company is to quarantine the offset until the review window has passed, which is unglamorous and will feel impossible when payroll is tight. Second, contemporaneous documentation is the whole game. The companies that lose are rarely the ones doing fake research, they are the ones who cannot prove real research two years after the fact, so write the technical record as the work happens rather than at claim time. Expect the July 2028 changes to push advisers toward conservative claiming well before they commence, and expect the pool of small claimants to shrink, which is the opposite of what an investment signal is supposed to do.